Re: RE24

[Pierangelo Masarati <ando@sys-net.it>]

Howard Chu wrote:
> Aaron Richton wrote:
>> If Ando's comment that Solaris poll() isn't working yet is true, I'd
>> like to see that work out. It wasn't clear to me whether he was
>> implicating slapd or Solaris, though...is there a next step here? (set
>> some autoconf to force poll() and make test, perhaps?)
> 
> No, there's no autoconf switch. You just need to make sure
> SLAP_X_DEVPOLL is defined when slapd/daemon.c is compiled, and then be
> prepared for sparks.
> 
>> On Thu, 1 Feb 2007, Howard Chu wrote:
>>
>>> OK, if no new issues are reported against HEAD in the next 24 hours
>>> then I think tomorrow we should kick out a new 2.4alpha.
>>>
>>> Offhand it looks to me like syncrepl enhancements are the only thing
>>> missing for a Beta. Anyone else have a list of items they want to
>>> resolve for 2.4?
> 
> Note the wording - I didn't ask for items "you want resolved (by someone
> else)" - that list would be a lot longer. I'm polling for what any
> developers have in-progress that they would like to have included, or
> what they plan to get finished by then.
> 
> What I'd like resolved, but don't have time to get to yet:
>   cn=config support for all the remaining backends and overlays that
>     still need to be converted:
>   back-dnssrv - this has no config options, so all it needs is
>     an empty cf table to shut up the warning.
>   back-meta - depends on many other things, primarily librewrite

Apart from rewriting, back-meta shouldn't be a big deal; it just has
really a lot of statements, and shares a lot of them with back-ldap,
occasionally on a per-target basis.

>   back-null - oddly enough, this has a config option...
>   back-passwd - has 1 option
>   back-perl - complicated because it can pass arbitrary options
>     to the underlying perl module. We can still accomodate this
>     using something like a perl- prefix or a olcDbPerlConfig:
>     attribute, just like how we handle DB_CONFIG.
>   back-relay - depends on librewrite / slapo-rwm
>   back-shell - easy enough
>   back-sql - haven't really looked

This should be trivial, there is no ordering issues, just a bunch of
statements.

> It looks like the majority of the overlays have been converted. rwm is
> still a problem.

I think the real problem is that shortcuts like "suffixmassage <DN>
[<DN>]" actually are converted to a bunch of librewrite statements which
may make little sense to those who don't like configuring librewrite
directly.  Or, if someone mixes up suffixmassage and direct statements,
things may get weird.  A plain approach would be to just pile up those
statements in an X-ORDERED bunch of values; this is supposed to work as
expected.

> collect.c, dyngroup.c, retcode.c, seqmod.c are all
> trivial.
> 
> Manpages for the useful overlays that are missing them would be nice.
> I'm not so worried about collect.c or seqmod.c since they aren't
> actually meant to be used, they're only meant as coding examples.
> 
> On a side-note, when I was benchmarking the connection manager I found
> it useful to doctor up back-null to actually return an entry in response
> to search requests. If anyone else would find this useful, I can commit
> that patch.

For a similar need, I typically use back-null(5) with slapo-retcode(5),
configured to return an entry with successful/error response code.

p.